ERP/ERF preprocessing
ERPERF prepares continuous recordings, corrects ocular artefacts, extracts event-related epochs, assesses artefacts and writes condition averages and optional clean-trial outputs. This guide describes the implemented EEG workflow, including BioSemi BDF and EGI RAW input; the historical ERP/ERF name is not a claim that every MEG system is supported.
Before processing
- Select a study folder and the recording’s format.
- Run MARKER for each recording, or supply an equivalent timing file. Select the prepared timing tag in ERPERF. Event recoding/subdivision belongs in timing preparation, not in epoch extraction.
- Check the sensor configuration, reference weights and recorded sample rate.
- Configure ocular correction, standard filters, epochs, artefact parameters and averaging outputs.
- Choose the processing branch, step range, interaction mode and reuse policy. Save a workspace and complete one representative file before a large batch.
ERPERF does not resample recordings automatically. Its configured rate must match the input before temporal filtering or reuse. If resampling is required, use Tools > Raw-data tools > Reduce and decimate first, and configure the filters for the resulting recording.
Processing order with ICA
| Step | Work |
|---|---|
| 1 | Prepare continuous data: transpose and rereference |
| 2 | Prepare ICA training rejection decisions |
| 3 | Decide ICA training rejection |
| 4 | Calculate ICA decomposition (Infomax) |
| 5 | Decide ICA component rejection |
| 6 | Apply ICA correction to continuous data |
| 7 | Filter continuous data for ERP/ERF analysis |
| 8 | Extract event-related epochs |
| 9 | Calculate statistical artefact parameters |
| 10 | Decide statistical artefact thresholds |
| 11 | Finish averaging and results |
ICA-training filters affect a temporary training copy. Component subtraction uses the minimally processed broadband recording. Standard ERP/ERF filters are applied after correction, before experimental epochs are extracted.
With ICA disabled, its training/correction steps are absent and the visible list is shorter. Regression-based ocular correction, when selected, operates on extracted epochs. Use step names, not remembered numbers, when changing correction methods.
Read by task
- Prepare inputs and open a recording
- Continuous preparation, rereferencing and filtering
- Ocular correction: ICA and regression
- Extract epochs and set their baseline
- Calculate epoch statistics and channel proposals
- Decide artefact thresholds and average trials
- Flow control, review passes and reuse
- Processing branches and subject overrides
- Group conditions and select accepted trials
- Processing Outcomes and Retained Trials
- Files, cleanup and troubleshooting
Working settings and persistence are explained in Analysis settings, workspaces and study folders.